Corn Leaf Disease Detection and Classification using Deep Learning Models
Abstract
Corn diseases are a major threat to the food security of the world. However, the existing deep learning models are "heavy" in terms of computation and are not feasible for mobile applications. This paper proposes a model by modifying the existing DenseNet and EfficientNet models for the classification of four different types of corn leaves, namely Common Rust, Cercospora Leaf Spot, Northern Leaf Blight, and Healthy. Using the PlantVillage dataset and the existing data augmentation strategy, the proposed model integrating EfficientNet and DenseNet models has shown a high accuracy in terms of validation. This model has shown the feasibility of the high-precision corn leaf disease detection with minimal memory usage. Experimental results demonstrate that the proposed hybrid EfficientNetB0–DenseNet121 model achieves a classification accuracy of 96.94% with reliable and consistent performance. The model provides an efficient solution for accurate corn leaf disease detection while maintaining low computational complexity. These results highlight its potential for practical deployment in intelligent and precision agriculture applications.